# load packages
library(tidyverse)
library(tidytext)
library(ngram)
library(igraph)
library(ggraph)
library(patchwork)
# library(svglite)
# library(geomnet)
# library(rgexf)
# library(qgraph)
# library(genBaRcode)
# library(ggraph)
library(ggiraph)
library(tidygraph)
library(scales)
#install.packages(
# "microViz",
#repos = c(davidbarnett = "https://david-barnett.r-universe.dev", getOption("repos"))
#)
#if (!require("BiocManager", quietly = TRUE))
# install.packages("BiocManager")
#BiocManager::install("phyloseq")
# library(microViz)
# library(tidygraph)
# library(plotly)
library(classInt)
library(grid)
# devtools::install_github("analyxcompany/ForceAtlas2")
library(ForceAtlas2)
# library(networkD3)A dynamic network approach to bilingual child data
About this document
This document contains the script used to analyze the data of two German-English bilingual children, “Fion” and “Silvie”, for the paper “A dynamic network approach to bilingual child data”. The method is loosely based on Ibbotson, Salnikov & Walker (2019). The data come from corpora described in more detail in e.g. Quick et al. (2018) and Koch, Endesfelder Quick & Hartmann (2025).
Data availability
The “Fion” data are already available on CHILDES. The “Silvie” data are currently in the process of being anonymized; we expect the data to become public in 2026.
Preliminaries
Loading some packages:
Toy example
This is just a toy example for illustrating the basic ideas:
testdf <- tibble(text = c("I love linguistics", "I like linguistics", "I love networks"))
test_graph <- testdf %>% unnest_tokens(output = "bigrams", input = "text", token = "ngrams", n = 2) %>% separate_wider_delim(cols = "bigrams", delim = " ", names = c("word1", "word2")) %>% graph_from_data_frame()
E(test_graph)$weight <- c(2,1,1,1,2,1)ggplot(layout) +
geom_edge_link(aes(x = x, y = y, xend = xend, yend = yend,
edge_width = weight), color = "gray") +
scale_edge_width(range = c(0.1, 0.7)) +
geom_node_point(col="red") +
geom_node_text(aes(label = name), position = position_nudge(y = .1)) +
scale_color_identity() + scale_fill_identity() +
theme_void() +
theme(legend.position = "none")
# ggsave("images/ilovenetworks.png", width = 7, height = 7, bg = "white")Data
CHI: Monolingual + mixed
# child data
d_fion <- read_csv("../../master/fion_CHI.csv")
d_silvie <- read_csv("../../master/silvie_CHI.csv")
# caregiver data
fion_cds <- read_csv("/Users/stefanhartmann/Library/CloudStorage/Dropbox/Input_Project/Data/master/fion_input_with_language_tags.csv.zip")
silvie_cds <- read_csv("/Users/stefanhartmann/Library/CloudStorage/Dropbox/Input_Project/Data/master/silvie_input_with_language_tags.csv.zip")
# remove all CDS data with "xxx" or only consisting of "<trs>" because the language tagging is often off.
fion_cds <- fion_cds[-grep("xxx", fion_cds$Utterance),]
silvie_cds <- silvie_cds[-grep("xxx", silvie_cds$Utterance),]
fion_cds <- fion_cds[-grep("<tr.*", fion_cds$Utterance),]
silvie_cds <- silvie_cds[-grep("<tr.*", silvie_cds$Utterance),]Data wrangling
Child data
We only want to take multi-word units into account, hence we first filter them out:
# add wordcount
d_fion$wordcount <- sapply(1:nrow(d_fion),
function(i) wordcount(d_fion$Utterance_clean[i]))
d_silvie$wordcount <- sapply(1:nrow(d_silvie),
function(i) wordcount(d_silvie$Utterance_clean[i]))
# only multi-word units
mwu_fion <- filter(d_fion, wordcount > 1)
mwu_silvie <- filter(d_silvie, wordcount > 1)Adding language tags: The column “Lang_Tags” already contains word-by-word tags for the code-mixed utterances, but not for the monolingual ones. However, the “type” column contains the information whether the utterance is English, German, or mixed. Hence for all non-mixed utterances, we can fill up the Lang_Tags column with this information. Before doing so, we add a few annotations that are missing in the files:
mwu_fion <- mwu_fion %>% mutate(Lang_Tags = case_when(
Utterance_clean == "darf ich this nicht aufraeumen" ~ "g g e g g",
Utterance_clean == "komm ich this nicht aufraeumen" ~ "g g e g g",
Utterance_clean == "und this my one" ~ "g e e e",
Utterance_clean == "nein no" ~ "g e",
Utterance_clean == "you did birthday in juni" ~ "e e e eg g",
Utterance_clean == "that istis ein birthday my nanny" ~ "e ge g e e e",
Utterance_clean == "und das ist von bob the builder und wendy" ~ "g g g g e e e g eg",
.default = Lang_Tags
))
mwu_silvie <- mwu_silvie %>% mutate(Lang_Tags = case_when(Utterance_clean == "und this noch" ~ "g e g",
Utterance_clean == "der postman pat" ~ "g e e",
Utterance_clean == "ja a game" ~ "g e e",
Utterance_clean == "ja this big one" ~ "g e e e",
Utterance_clean == "ja a starfish" ~ "g e e",
Utterance_clean == "ja the ribbon" ~ "g e e",
Utterance_clean == "das heisst naemlich train train train train" ~ "g g g e e e e",
.default = Lang_Tags))Also, there are some inconsistencies in the tagging that lead to more factor levels than necessary, we correct those:
mwu_fion$Lang_Tags <- gsub("e\\(meta\\)", "e", mwu_fion$Lang_Tags)
mwu_fion$Lang_Tags <- gsub("ge", "eg", mwu_fion$Lang_Tags)
mwu_fion$Lang_Tags <- gsub("m", "eg", mwu_fion$Lang_Tags)
mwu_silvie$Lang_Tags <- gsub("ge", "eg", mwu_silvie$Lang_Tags)
mwu_silvie$Lang_Tags <- gsub("m", "eg", mwu_silvie$Lang_Tags)Now we can proceed:
# add language tags
mwu_fion$Lang_Tags <- gsub("[[:punct:]]", "", mwu_fion$Lang_Tags)
mwu_silvie$Lang_Tags <- gsub("[[:punct:]]", "", mwu_silvie$Lang_Tags)
# add language tags on a word-by-word-basis to the non-code-mixed utterances
mwu_fion$Lang_Tags <- sapply(1:nrow(mwu_fion), function(i) ifelse(is.na(mwu_fion$Lang_Tags[i]), ifelse(mwu_fion[i,]$type=="german", paste0(rep("g", mwu_fion[i,]$wordcount), collapse = " "), paste0(rep("e", mwu_fion[i,]$wordcount), collapse = " ")), mwu_fion$Lang_Tags[i]))
mwu_silvie$Lang_Tags <- sapply(1:nrow(mwu_silvie), function(i) ifelse(is.na(mwu_silvie$Lang_Tags[i]), ifelse(mwu_silvie[i,]$type=="german", paste0(rep("g", mwu_silvie[i,]$wordcount), collapse = " "), paste0(rep("e", mwu_silvie[i,]$wordcount), collapse = " ")), mwu_silvie$Lang_Tags[i]))Get bigrams
Child data
As we are interested in transition probabilities between words, we need bigrams, which we get using the unnest_tokens function from the tidytext package. For each bigram, we also want the language information for the individual words, which is why we also extract bigrams from the Lang_Tags column in a second step and then join the dataframes. Finally, we split up the bigrams so that word1 and word2 are in different columns, which makes it easier to calculate the transition probabilities.
bigrams_fion <- mwu_fion %>% unnest_tokens(bigram, Utterance_clean, token = "ngrams", n = 2, drop = FALSE)
bigrams_silvie <- mwu_silvie %>% unnest_tokens(bigram, Utterance_clean, token = "ngrams", n = 2, drop = FALSE)
bigrams_fion <- bind_cols(bigrams_fion,
mwu_fion %>% unnest_tokens(bigram_LangTag, Lang_Tags, token = "ngrams", n = 2, drop = FALSE) %>% select(bigram_LangTag))
bigrams_silvie <- bind_cols(bigrams_silvie,
mwu_silvie %>% unnest_tokens(bigram_LangTag, Lang_Tags, token = "ngrams", n = 2, drop = FALSE) %>% select(bigram_LangTag))
# one column for each word
bigrams_fion <- bigrams_fion %>% separate(bigram, c("word1", "word2"), sep = " ", remove = F)
bigrams_silvie <- bigrams_silvie %>% separate(bigram, c("word1", "word2"), sep = " ", remove = F)
bigrams_fion <- bigrams_fion %>% separate(bigram_LangTag, c("LangTag1", "LangTag2"), sep = " ", remove = F)
bigrams_silvie <- bigrams_silvie %>% separate(bigram_LangTag, c("LangTag1", "LangTag2"), sep = " ", remove = F)
# add child column
bigrams_fion <- mutate(bigrams_fion, Child = "Fion")
bigrams_silvie <- mutate(bigrams_silvie, Child = "Silvie")CDS data
As in the child data, we extract bigrams:
# get bigrams --------------------------------------------------
bigrams_cds_fion <- fion_cds %>% unnest_tokens(bigram, Utterance_clean, token = "ngrams", n = 2)
bigrams_cds_silvie <- silvie_cds %>% unnest_tokens(bigram, Utterance_clean, token = "ngrams", n = 2)
# one column for each word
bigrams_cds_fion <- bigrams_cds_fion %>% separate(bigram, c("word1", "word2"), sep = " ", remove = F)
bigrams_cds_silvie <- bigrams_cds_silvie %>% separate(bigram, c("word1", "word2"), sep = " ", remove = F)Transition probabilities
As we are interested in the transition probabilities between words, we add those to the dataframes.
get_transition_probabilities <- function(df_bigrams, input_column = "Utterance_clean", age_range = "all", Speaker = "all") {
# filter
if(any(age_range!="all")) {
#df_orig <- filter(df_orig, age_range %in% age_range)
df_bigrams <- filter(df_bigrams, age_range %in% age_range)
}
if(any(Speaker!="all")) {
#df_orig <- filter(df_orig, Speaker %in% Speaker)
df_bigrams <- filter(df_bigrams, Speaker %in% Speaker)
}
# unigrams
unigrams <- df_bigrams %>% select(Utt_no, all_of(input_column)) %>% unique() %>% unnest_tokens(output = "unigram", input = input_column, token = "ngrams", n = 1)
# unigrams <- df_orig %>% unnest_tokens(output = "unigram", input = input_column, token = "ngrams", n = 1)
unigrams_tbl <- unigrams$unigram %>% table() %>% as.data.frame() %>% setNames(c("unigram", "Freq"))
# count co-occurrence frequencies of bigrams:
# first grouped by Utterance number so that
# the bigrams do not cross utterance boundaries,
# then summing up across utterances.
bigrams_df_tbl <- df_bigrams %>% group_by(Utt_no, word1, word2) %>% summarise(
n = n()
) %>% na.omit() %>% group_by(word1, word2) %>%
summarise(
n = sum(n)
)
# add unigram frequencies
bigrams_df_tbl <- left_join(bigrams_df_tbl, unigrams_tbl, by = c("word1" = "unigram")) %>% setNames(c("word1", "word2", "n", "n_word1"))
bigrams_df_tbl <- left_join(bigrams_df_tbl, unigrams_tbl, by = c("word2" = "unigram")) %>% setNames(c("word1", "word2", "n", "n_word1", "n_word2"))
# add backward and forward transitional probabilities
bigrams_df_tbl <- bigrams_df_tbl %>% mutate(ftp = n / n_word1,
btp = n / n_word2)
# return
return(bigrams_df_tbl)
}Function for periodization
This function adds a `Months` column to the data in which several months can be binned into larger groups. This makes it easier to try out different periodization options (or to stick with the original months data, i.e. have one network per month).
To make an informed decision about the way the data are split up into time slices, let’s first take a quick look at the distribution of data across the timespan:
Child speech
# number of utterances
d_fion %>% group_by(Month) %>% summarise(
n_utterances = n(),
n_files = length(unique(Filename))
) %>% ggplot(aes(x=Month, y = n_utterances, label = n_files)) +
geom_point() +
geom_line(group = 1) +
geom_text(position = position_stack(), vjust = -0.4) +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
ylab("Number of utterances") +
ggtitle("Fion (numbers indicate number of transcripts)") + d_silvie %>% group_by(Month) %>% summarise(
n_utterances = n(),
n_files = length(unique(Filename))
) %>% ggplot(aes(x=Month, y = n_utterances, label = n_files)) +
geom_point() +
geom_line(group = 1) +
geom_text(position = position_stack(), vjust = -0.4) +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
ylab("Number of utterances") +
ggtitle("Silvie (numbers indicate number of transcripts)")
# number of words
d_fion %>% group_by(Month) %>% summarise(
n_words = sum(Wordcount),
n_files = length(unique(Filename))
) %>% ggplot(aes(x=Month, y = n_words, label = n_files)) +
geom_point() +
geom_line(group = 1) +
geom_text(position = position_stack(), vjust = -0.4) +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
ylab("Number of words") +
ggtitle("Fion (numbers indicate number of transcripts)") + d_silvie %>% group_by(Month) %>% summarise(
n_words = sum(Wordcount),
n_files = length(unique(Filename))
) %>% ggplot(aes(x=Month, y = n_words, label = n_files)) +
geom_point() +
geom_line(group = 1) +
geom_text(position = position_stack(), vjust = -0.4) +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
ylab("Number of words") +
ggtitle("Silvie (numbers indicate number of transcripts)")
In both datasets, we have fewer transcripts and, hence, fewer words in the later periods. Thus, it seems useful to work with three-month periods and samples of 450 utterances.
# function for getting equal-sized intervals
get_equal_bins <- function(x, n) {
cur_var <- 1:length(unique(x))
cur_breaks <- round(seq(1, length(unique(x)), by = length(unique(x)) / n))
# add last number to breaks
cur_breaks[length(cur_breaks)+1] <- cur_var[length(cur_var)]
# add 0 to var and breaks so that we can calculate +1
# below to avoid overlaps between two groups
cur_var <- c(0, cur_var)
cur_breaks[1] <- 0
# cur_breaks[1] <- 0 # to make sure that it starts with 1
cur_list <- lapply(1:(length(cur_breaks)-1), function(i) cur_var[((which(cur_var==cur_breaks[i])+1)):(which(cur_var==cur_breaks[1+i]))])
return(cur_list)
}
# function for adding months column
add_month_bins <- function(df, n_bins) {
cur_bins <- get_equal_bins(df$Month, n_bins)
# list to dataframe
cur_bins_df <- do.call(rbind, lapply(1:length(cur_bins), function(i) tibble(index = i,
no = cur_bins[[i]])))
# get bin range
bins_tbl01 <- tibble(age = unique(df$Month),
no = 1:length(unique(df$Month)))
# join
cur_bins_df <- left_join(cur_bins_df, bins_tbl01)
# bins in character form (from a to b)
cur_bins_ch <- sapply(1:length(cur_bins), function(i) paste0(unique(df$Month)[cur_bins[[i]][1]], "-",
unique(df$Month)[cur_bins[[i]][length(cur_bins[[i]])]]))
# in tabular form
bins_tbl <- tibble(bin = 1:length(cur_bins),
age_range = cur_bins_ch)
# add to existing dataframe
cur_bins_df <- left_join(cur_bins_df, bins_tbl, by = c("index" = "bin"))
# return Months column
cur_df_with_age_range <- left_join(df, select(cur_bins_df, age, age_range), by = c("Month" = "age"))
return(cur_df_with_age_range)
}Child-directed speech
Now we turn to the child-directed speech data:
# count:
# first grouped by Utterance number so that
# the bigrams do not cross utterance boundaries,
# then summing up across utterances.
bigrams_fion_cds_count <- bigrams_cds_fion %>% group_by(Utt_no, word1, word2, lang) %>% summarise(
n = n()
) %>% na.omit() %>% group_by(word1, word2, lang) %>%
summarise(
n = sum(n)
) %>% arrange(desc(n))
bigrams_silvie_cds_count <- bigrams_cds_silvie %>% group_by(Utt_no, word1, word2, lang) %>% summarise(
n = n()
) %>% na.omit() %>% group_by(word1, word2, lang) %>%
summarise(
n = sum(n)
) %>% arrange(desc(n))Getting periods and samples
We want to work with three-month periods, hence we divide the Fion data into 21/3 = 7 bins, the Silvie data into 18/3 = 6 bins. For the child-directed speech data, we do not expect as significant differences between the age spans, so we only work with ~six-month bins.
Periods - child data
# how many months for Fion and Silvie?
length(unique(d_fion$Month))[1] 21length(unique(d_silvie$Month))[1] 18# add bins
bigrams_fion <- bigrams_fion %>% add_month_bins(n_bins = 7)
bigrams_silvie <- bigrams_silvie %>% add_month_bins(n_bins = 6)
d_fion <- d_fion %>% add_month_bins(n_bins = 7)
d_silvie <- d_silvie %>% add_month_bins(n_bins = 6)Periods - CDS data
# Get bins
bigrams_cds_fion <- bigrams_cds_fion %>% add_month_bins(n = 3)
bigrams_cds_silvie <- bigrams_cds_silvie %>% add_month_bins(n = 3)
fion_cds <- fion_cds %>% add_month_bins(n = 3)
silvie_cds <- silvie_cds %>% add_month_bins(n = 3)Sampling - Child data
The following code compiles 450-utterance samples from three-month periods for the child data, and 45,000-utterance samples from the child-directed speech data for each ~6-month period.
# get samples
set.seed(1234567)
# Fion:
for(i in 1:7) {
cur_fion <- filter(bigrams_fion, age_range == levels(factor(bigrams_fion$age_range))[i])
set.seed(i)
cur_samples <- sample(1:length(unique(cur_fion$Utt_no)), 450)
# cur_samples <- lapply(1:100, function(i) sample(1:length(unique(cur_fion$Utt_no)), 450)) # alternative with bootstrapping
assign(paste0("bigrams_fion0", i), cur_fion[unlist(cur_samples),])
}
# Silvie:
for(i in 1:6) {
cur_silvie <- filter(bigrams_silvie, age_range == levels(factor(bigrams_silvie$age_range))[i])
set.seed(i)
cur_samples <- sample(1:length(unique(cur_silvie$Utt_no)), 450)
#cur_samples <- lapply(1:100, function(i) sample(1:length(unique(cur_silvie$Utt_no)), 450))
assign(paste0("bigrams_silvie0", i), cur_silvie[unlist(cur_samples),])
}Sampling - CDS data
# backup
bigrams_cds_fion_with_unigrams <- bigrams_cds_fion
bigrams_cds_silvie_with_unigrams <- bigrams_cds_silvie
# omit utterances consisting of only one word
bigrams_cds_fion <- bigrams_cds_fion %>% filter(Wordcount > 1)
bigrams_cds_silvie <- bigrams_cds_silvie %>% filter(Wordcount > 1)
# get 25,000-utterance samples for each of the bins
set.seed(12345)
for(i in 1:3) {
cur_bigrams <- bigrams_cds_fion %>% filter(age_range == levels(factor(bigrams_cds_fion$age_range))[i])
cur_spl <- sample(1:nrow(cur_bigrams), 45000)
assign(paste0("bigrams_cds_fion0", i), cur_bigrams[cur_spl,])
}
set.seed(12345)
for(i in 1:3) {
cur_bigrams <- bigrams_cds_silvie %>% filter(age_range == levels(factor(bigrams_cds_silvie$age_range))[i])
cur_spl <- sample(1:nrow(cur_bigrams), 45000)
assign(paste0("bigrams_cds_silvie0", i), cur_bigrams[cur_spl,])
}Getting networks
The following function calculates the actual networks by calculating transition probabilities, filtering out all instances attested less than n_min (default: 5) times.
Networks for child language
get_network <- function(bigram_df,n_min = 0, modularity_measure = FALSE, edges = "ftp") {
# count the bigram_df
bigrams_count <- bigram_df %>% group_by(LangTag1, LangTag2) %>% count(word1, word2, sort = T)
# filter out all below n_min
l <- bigrams_count %>%
filter(n >= n_min)
# add transitional probabilities
l1 <- left_join(l, select(get_transition_probabilities(bigram_df, input_column = "Utterance_clean"), -n), by = c("word1", "word2") )
# check if there are data
if(nrow(l1) > 0) {
# get bigram graph edges and vertices
bigram_graph <- l1 %>%
filter(n >= n_min) %>%
ungroup %>% select(word1, word2, all_of(edges)) %>% graph_from_data_frame(directed = FALSE)
# set weight attributes
if(edges == "ftp") {
bigram_graph <- set_edge_attr(bigram_graph, "weight", value = l1$ftp)
} else if(edges=="btp") {
bigram_graph <- set_edge_attr(bigram_graph, "weight", value = l1$btp)
} else if(edges=="n") {
bigram_graph <- set_edge_attr(bigram_graph, "weight", value = l1$n)
}
# set labels
V(bigram_graph)$label <- V(bigram_graph)$name
# Louvain clustering
lv <- cluster_louvain(bigram_graph, resolution = .3)
# add Louvain clustering to graph
V(bigram_graph)$community <- membership(lv)
# add color for each community
cur_palette <- viridis::cividis(n = max(V(bigram_graph)$community))
cur_communities <- V(bigram_graph)$community
V(bigram_graph)$community_color <- sapply(1:length(V(bigram_graph)$community), function(i) cur_communities[i])
# add language and word frequency as attributes
# to do so, we need a list of unigrams
# language tags of unigrams
unigram_LangTags <- bind_cols(bigram_df %>% select(Utt_no, Utterance_clean, Lang_Tags) %>% unique() %>% unnest_tokens(output = "unigram", input = "Utterance_clean", token = "ngrams", n = 1),
select(bigram_df %>% select(Utt_no, Utterance_clean, Lang_Tags) %>% unique() %>% unnest_tokens(output = "unigram_LangTag", input = Lang_Tags, token = "ngrams", n = 1), unigram_LangTag)) %>% select(unigram, unigram_LangTag) %>% unique()
# get frequencies of individual words
unigrams_freqs <- bigram_df %>% select(Utt_no, Utterance_clean) %>% unique() %>% unnest_tokens(output = "unigram", input = Utterance_clean) %>% group_by(unigram) %>% summarise(
n = n()
)
# add unigram LangTags as attributes to the graph
V(bigram_graph)$language <- sapply(1:length(V(bigram_graph)), function(i) unigram_LangTags[which(unigram_LangTags$unigram == V(bigram_graph)$name[i]),]$unigram_LangTag[1])
# add color as attribute
V(bigram_graph)$color <- case_when(V(bigram_graph)$language == "g" ~ "salmon",
V(bigram_graph)$language == "e" ~ "deepskyblue",
V(bigram_graph)$language == "eg" ~ "tan",
.default = "grey")
# add frequency as attribute
V(bigram_graph)$Freq <- sapply(1:length(V(bigram_graph)), function(i) unigrams_freqs[which(unigrams_freqs$unigram==V(bigram_graph)$name[i]),]$n[1])
# return graph or modularity measure
if(modularity_measure) {
return(modularity(lv))
} else {
return(bigram_graph)
}
}
}Networks for child-directed speech
# add missing columns (for compatibility with the above-defined functions)
bigrams_cds_fion$LangTag1 <- bigrams_cds_fion$lang
bigrams_cds_fion$LangTag2 <- bigrams_cds_fion$lang
bigrams_cds_fion$Lang_Tags <- bigrams_cds_fion$lang
bigrams_cds_fion$Utterance_clean <- bigrams_cds_fion$Utterance
bigrams_cds_silvie$LangTag1 <- bigrams_cds_silvie$lang
bigrams_cds_silvie$LangTag2 <- bigrams_cds_silvie$lang
bigrams_cds_silvie$Lang_Tags <- bigrams_cds_silvie$lang
bigrams_cds_silvie$Utterance_clean <- bigrams_cds_silvie$Utterance
bigrams_cds_fion01$LangTag1 <- bigrams_cds_fion01$lang
bigrams_cds_fion01$LangTag2 <- bigrams_cds_fion01$lang
bigrams_cds_fion01$Lang_Tags <- bigrams_cds_fion01$lang
bigrams_cds_fion01$Utterance_clean <- bigrams_cds_fion01$Utterance
bigrams_cds_silvie01$LangTag1 <- bigrams_cds_silvie01$lang
bigrams_cds_silvie01$LangTag2 <- bigrams_cds_silvie01$lang
bigrams_cds_silvie01$Lang_Tags <- bigrams_cds_silvie01$lang
bigrams_cds_silvie01$Utterance_clean <- bigrams_cds_silvie01$Utterance
bigrams_cds_fion02$LangTag1 <- bigrams_cds_fion02$lang
bigrams_cds_fion02$LangTag2 <- bigrams_cds_fion02$lang
bigrams_cds_fion02$Lang_Tags <- bigrams_cds_fion02$lang
bigrams_cds_fion02$Utterance_clean <- bigrams_cds_fion02$Utterance
bigrams_cds_silvie02$LangTag1 <- bigrams_cds_silvie02$lang
bigrams_cds_silvie02$LangTag2 <- bigrams_cds_silvie02$lang
bigrams_cds_silvie02$Lang_Tags <- bigrams_cds_silvie02$lang
bigrams_cds_silvie02$Utterance_clean <- bigrams_cds_silvie02$Utterance
bigrams_cds_fion03$LangTag1 <- bigrams_cds_fion03$lang
bigrams_cds_fion03$LangTag2 <- bigrams_cds_fion03$lang
bigrams_cds_fion03$Lang_Tags <- bigrams_cds_fion03$lang
bigrams_cds_fion03$Utterance_clean <- bigrams_cds_fion03$Utterance
bigrams_cds_silvie03$LangTag1 <- bigrams_cds_silvie03$lang
bigrams_cds_silvie03$LangTag2 <- bigrams_cds_silvie03$lang
bigrams_cds_silvie03$Lang_Tags <- bigrams_cds_silvie03$lang
bigrams_cds_silvie03$Utterance_clean <- bigrams_cds_silvie03$Utterance
# function for getting CDS networks
get_network_cds <- function(bigram_df, n_min = 0, modularity_measure = FALSE, edges = "ftp") {
bigrams_count <- bigram_df %>% group_by(LangTag1, LangTag2) %>% count(word1, word2, sort = T)
# bigrams_count <- na.omit(bigrams_count)
# filter out all below n_min
l <- bigrams_count %>%
filter(n >= n_min)
# add transitional probabilities
l1 <- left_join(l, select(get_transition_probabilities(bigram_df, input_column = "Utterance_clean"), -n), by = c("word1", "word2") )
# omit NAs (usually blank spaces)
l1 <- na.omit(l1)
# check if there are data
if(nrow(l1) > 0) {
# get bigram graph edges and vertices
bigram_graph <- l1 %>%
filter(n >= n_min) %>%
ungroup %>% select(word1, word2, all_of(edges)) %>% graph_from_data_frame(directed = FALSE)
# set weight attributes
if(edges == "ftp") {
bigram_graph <- set_edge_attr(bigram_graph, "weight", value = l1$ftp)
} else if (edges == "btp") {
bigram_graph <- set_edge_attr(bigram_graph, "weight", value = l1$btp)
} else if(edges == "n") {
bigram_graph <- set_edge_attr(bigram_graph, "weight", value = l1$n)
}
# set labels
V(bigram_graph)$label <- V(bigram_graph)$name
# Louvain clustering
lv <- cluster_louvain(bigram_graph, resolution = .3)
# add Louvain clustering to graph
V(bigram_graph)$community <- membership(lv)
# add color for each community
cur_palette <- viridis::cividis(n = max(V(bigram_graph)$community))
cur_communities <- V(bigram_graph)$community
V(bigram_graph)$community_color <- sapply(1:length(V(bigram_graph)$community), function(i) cur_communities[i])
# add language and word frequency as attributes
# to do so, we need a list of unigrams
# language tags of unigrams
lang_tbl <- unique(select(bigram_df, word1, word2, lang)) %>% pivot_longer(cols = 1:2) %>% select(value, lang) %>% setNames(c("word", "language")) %>% unique()
# add language tags to network
V(bigram_graph)$language <- sapply(1:length(V(bigram_graph)$name), function(i) lang_tbl[which(lang_tbl$word == V(bigram_graph)$name[i])[1],]$language)
# get frequencies of individual words
unigrams_freqs <- bigram_df %>% select(Utt_no, Utterance) %>% unique() %>% unnest_tokens(output = "unigram", input = Utterance) %>% group_by(unigram) %>% summarise(
n = n()
)
# add color as attribute
V(bigram_graph)$color <- case_when(V(bigram_graph)$language == "de" ~ "salmon",
V(bigram_graph)$language == "en" ~ "deepskyblue",
V(bigram_graph)$language == "mixed" ~ "tan",
.default = "grey")
# add frequency as attribute
V(bigram_graph)$Freq <- sapply(1:length(V(bigram_graph)), function(i) unigrams_freqs[which(unigrams_freqs$unigram==V(bigram_graph)$name[i]),]$n[1])
}
# return graph or modularity measure
if(modularity_measure) {
return(modularity(lv))
} else {
return(bigram_graph)
}
}Visualization
The functions created above can now be combined to create networks for different age spans.
# function for plotting
get_plot <- function(cur_network, myseed = 1999, min_freq = 0, interactive = FALSE, repel = TRUE, max.overlaps = 20, communities = "all", input = "igraph", ellipse_alpha = 0.1, layout = "ForceAtlas2") {
if(input == "layout") {
layout <- cur_network
} else {
# get plot layout
layout <- create_layout(cur_network, layout = "fr")
}
if(any(layout == "fr")) {
layout <- layout
} else if(any(layout == "ForceAtlas2")) {
set.seed(myseed)
layout <- create_layout(test_graph, layout = "fr")
layout[,1:2] <- layout.forceatlas2(test_graph, directed = TRUE, iterations = 1000, plotstep = 10) %>% as.data.frame() %>% setNames(c("x", "y"))
}
# allow for selecting individual communities
if(any(communities != "all")) {
layout <- filter(layout, community %in% communities)
}
# Build plot
set.seed(myseed)
p <- ggplot(layout) +
geom_edge_link(aes(x = x, y = y, xend = xend, yend = yend,
edge_width = weight, alpha = weight), color = "gray") +
scale_edge_width(range = c(0.1, 0.5)) +
geom_point_interactive(
aes(x = x, y = y, tooltip = name, color = color, size = Freq)
) +
geom_node_text(aes(label = ifelse(Freq > min_freq, name, ""), size = Freq, color = community_color), repel = repel, max.overlaps = max.overlaps) +
stat_ellipse(aes(x=x, y=y, group = as.factor(community), fill = community_color),
geom = "polygon", alpha = ellipse_alpha, color = NA) +
scale_color_identity() + scale_fill_identity() +
theme_void() +
theme(legend.position = "none")
if(interactive) {
# Zoomable plot with girafe
g <- girafe(
ggobj = p,
options = list(
opts_zoom(min = 1, max = 60),
opts_toolbar(saveaspng = TRUE)
)
)
return(g)
} else {
return(p)
}
}Plots for publication
Child speech
Overall networks
# create seven objects with Fion's networks
for(i in 1:7) {
assign(paste0("network_fion0", i), get(paste0("bigrams_fion0", i)) %>% get_network())
}
# create six objects with Silvie's networks
for(i in 1:6) {
assign(paste0("network_silvie0", i), get(paste0("bigrams_silvie0", i)) %>% get_network())
}
# create seven plot objects named p_f1 to p_f7 for Fion's networks
for(i in 1:7) {
assign(paste0("p_f", i), get(paste0("network_fion0", i)) %>% get_plot(repel = TRUE, max.overlaps = 50, min_freq = 5) + ggtitle(levels(factor(bigrams_fion$age_range))[i]) + theme(plot.title = element_text(face = "bold", hjust = 0.5)))
}
# create six plot objects named p_f1 to p_f7 for Silvie's networks
for(i in 1:6) {
assign(paste0("p_s", i), get(paste0("network_silvie0", i)) %>% get_plot(repel = TRUE, min_freq = 5, max.overlaps = 50) + ggtitle(levels(factor(bigrams_silvie$age_range))[i]) + theme(plot.title = element_text(face = "bold", hjust = 0.5)))
}
# combine all in one big plot
# titles:
p_t1 <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Bigram', ' network, Fion'))", size = 12, parse = TRUE) + theme_void()
p_t2 <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Bigram', ' network, Silvie'))", size = 12, parse = TRUE) + theme_void()
# composite plot:
p_t1 /
(p_f1 | p_f2 | p_f3) /
(p_f4 | p_f5 | p_f6) /
p_f7 /
p_t2 /
(p_s1 | p_s2 | p_s3) /
(p_s4 | p_s5 | p_s6) +
plot_layout(heights = c(.2,1,1,1,.2,1,1))
ggsave("images/fion_silvie_networks450.png", width = 15, height = 20.5)Largest communities
# get largest communities - Fion
for(i in 1:7) {
cur_network <- get_network(get(paste0("bigrams_fion0", i)), edges = "n")
cur_network_layout <- create_layout(cur_network, layout = "fr")
largest_communities <- cur_network_layout %>% group_by(community) %>% summarise(
n = n()
) %>% arrange(desc(n)) %>% head(4) %>% select(community) %>% as.vector() %>% unname() %>% unlist()
prop_top_communities <- table(cur_network_layout$community) %>% prop.table() %>% as_tibble(.name_repair="unique") %>% setNames(c("community", "n")) %>% arrange(desc(n)) %>% head(4)
p0001 <- cur_network %>% get_plot(communities=largest_communities[1]) + ggtitle(paste0("Community ", 1, " (", paste0(round(prop_top_communities$n[1], digits = 4)*100, "%"), ")"))
p0002 <- cur_network %>% get_plot(communities=largest_communities[2]) + ggtitle(paste0("Community ", 2, " (", paste0(round(prop_top_communities$n[2], digits = 4)*100, "%"), ")"))
p0003 <- cur_network %>% get_plot(communities=largest_communities[3]) + ggtitle(paste0("Community ", 3, " (", paste0(round(prop_top_communities$n[3], digits = 4)*100, "%"), ")"))
p0004 <- cur_network %>% get_plot(communities=largest_communities[4]) + ggtitle(paste0("Community ", 4, " (", paste0(round(prop_top_communities$n[4], digits = 4)*100, "%"), ")"))
cur_p <- (p0001 | p0002 | p0003 | p0004)
assign(paste0("p_lc_f_period", i), cur_p)
}
# get largest communities - Silvie
for(i in 1:6) {
cur_network <- get_network(get(paste0("bigrams_silvie0", i)), edges = "n")
cur_network_layout <- create_layout(cur_network, layout = "fr")
largest_communities <- cur_network_layout %>% group_by(community) %>% summarise(
n = n()
) %>% arrange(desc(n)) %>% head(4) %>% select(community) %>% as.vector() %>% unname() %>% unlist()
prop_top_communities <- table(cur_network_layout$community) %>% prop.table() %>% as_tibble(.name_repair="unique") %>% setNames(c("community", "n")) %>% arrange(desc(n)) %>% head(4)
p0001 <- cur_network %>% get_plot(communities=largest_communities[1]) + ggtitle(paste0("Community ", 1, " (", paste0(round(prop_top_communities$n[1], digits = 4)*100, "%"), ")"))
p0002 <- cur_network %>% get_plot(communities=largest_communities[2]) + ggtitle(paste0("Community ", 2, " (", paste0(round(prop_top_communities$n[2], digits = 4)*100, "%"), ")"))
p0003 <- cur_network %>% get_plot(communities=largest_communities[3]) + ggtitle(paste0("Community ", 3, " (", paste0(round(prop_top_communities$n[3], digits = 4)*100, "%"), ")"))
p0004 <- cur_network %>% get_plot(communities=largest_communities[4]) + ggtitle(paste0("Community ", 4, " (", paste0(round(prop_top_communities$n[4], digits = 4)*100, "%"), ")"))
cur_p <- (p0001 | p0002 | p0003 | p0004)
assign(paste0("p_lc_s_period", i), cur_p)
}
# plot titles
p_t1 <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Largest', ' communities, Fion'))", size = 12, parse = TRUE) + theme_void()
for(i in 1:7) {
assign(paste0("title_agerange_fion", i), ggplot() + annotate("text", x = 4, y = 25, label = levels(factor(bigrams_fion$age_range))[i], size = 8, parse = FALSE) + theme_void())
}
p_t2 <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Largest', ' communities, Silvie'))", size = 12, parse = TRUE) + theme_void()
for(i in 1:6) {
assign(paste0("title_agerange_silvie", i), ggplot() + annotate("text", x = 4, y = 25, label = levels(factor(bigrams_fion$age_range))[i], size = 8, parse = FALSE) + theme_void())
}
# complete plot
p_t1 /
title_agerange_fion1 /
p_lc_f_period1 /
title_agerange_fion2 /
p_lc_f_period2 /
title_agerange_fion3 /
p_lc_f_period3 /
title_agerange_fion4 /
p_lc_f_period4 /
title_agerange_fion5 /
p_lc_f_period5 /
title_agerange_fion6 /
p_lc_f_period6 /
title_agerange_fion7 /
p_lc_f_period7 + plot_layout(heights = c(.2, rep(c(.1,1), 7)))
# ggsave("images/largest_communities_fion_new.png", width = 16, height = 25)
# plot titles
p_t1 <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Largest', ' communities, Silvie'))", size = 12, parse = TRUE) + theme_void()
for(i in 1:7) {
assign(paste0("title_agerange_silvie", i), ggplot() + annotate("text", x = 4, y = 25, label = levels(factor(bigrams_silvie$age_range))[i], size = 8, parse = FALSE) + theme_void())
}
p_t2 <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Largest', ' communities, Silvie'))", size = 12, parse = TRUE) + theme_void()
for(i in 1:6) {
assign(paste0("title_agerange_silvie", i), ggplot() + annotate("text", x = 4, y = 25, label = levels(factor(bigrams_silvie$age_range))[i], size = 8, parse = FALSE) + theme_void())
}
# complete plot
p_t2 /
title_agerange_silvie1 /
p_lc_s_period1 /
title_agerange_silvie2 /
p_lc_s_period2 /
title_agerange_silvie3 /
p_lc_s_period3 /
title_agerange_silvie4 /
p_lc_s_period4 /
title_agerange_silvie5 /
p_lc_s_period5 /
title_agerange_silvie6 /
p_lc_s_period6 + plot_layout(heights = c(.2, rep(c(.1,1), 6)))
ggsave("images/largest_communities_silvie_new.png", width = 16, height = 25)Child-directed speech
Overall networks
# networs for each of the three time slots, separately for mothers and fathers
# Silvie - mother
silvie_MOT01 <- bigrams_cds_silvie01 %>% filter(Speaker == "MOT") %>% get_network_cds()
silvie_MOT02 <- bigrams_cds_silvie02 %>% filter(Speaker == "MOT") %>% get_network_cds()
silvie_MOT03 <- bigrams_cds_silvie03 %>% filter(Speaker == "MOT") %>% get_network_cds()
# Silvie - father
silvie_FAT01 <- bigrams_cds_silvie01 %>% filter(Speaker == "FAT") %>% get_network_cds()
silvie_FAT02<- bigrams_cds_silvie02 %>% filter(Speaker == "FAT") %>% get_network_cds()
silvie_FAT03 <- bigrams_cds_silvie03 %>% filter(Speaker == "FAT") %>% get_network_cds()
# Fion - mother
fion_MOT01 <- bigrams_cds_fion01 %>% filter(Speaker == "MOT") %>% get_network_cds()
fion_MOT02 <- bigrams_cds_fion02 %>% filter(Speaker == "MOT") %>% get_network_cds()
fion_MOT03 <- bigrams_cds_fion03 %>% filter(Speaker == "MOT") %>% get_network_cds()
# Fion - father
fion_FAT01 <- bigrams_cds_fion01 %>% filter(Speaker == "FAT") %>% get_network_cds()
fion_FAT02<- bigrams_cds_fion02 %>% filter(Speaker == "FAT") %>% get_network_cds()
fion_FAT03 <- bigrams_cds_fion03 %>% filter(Speaker == "FAT") %>% get_network_cds()We don’t include the images here to keep the size of the file a bit lower, but the png files can be found in the “images” folder (`images/networks_cds.png`).
# plot networks
# titles
p_t1 <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Bigram', ' networks, Fion', '\\'s', ' mother'))", size = 16, parse = TRUE) + theme_void()
p_t2 <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Bigram', ' networks, Fion', '\\'s', ' father'))", size = 16, parse = TRUE) + theme_void()
p_t3 <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Bigram', ' networks, Silvie', '\\'s', ' mother'))", size = 16, parse = TRUE) + theme_void()
p_t4 <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Bigram', ' networks, Silvie', '\\'s', ' father'))", size = 16, parse = TRUE) + theme_void()
p_t1 /
(get_plot(fion_MOT01, repel = F, min_freq =3, ellipse_alpha = .001) |
get_plot(fion_MOT02, repel = F, min_freq =3, ellipse_alpha = .001) |
get_plot(fion_MOT03, repel = F, min_freq =3, ellipse_alpha = .001) ) /
p_t2 /
(get_plot(fion_FAT01, repel = F, min_freq =3, ellipse_alpha = .001) |
get_plot(fion_FAT02, repel = F, min_freq =3, ellipse_alpha = .001) |
get_plot(fion_FAT03, repel = F, min_freq =3, ellipse_alpha = .001) ) /
p_t3 /
(get_plot(silvie_FAT01, repel = F, min_freq =3, ellipse_alpha = .001) |
get_plot(silvie_FAT02, repel = F, min_freq =3, ellipse_alpha = .001) |
get_plot(silvie_FAT03, repel = F, min_freq =3, ellipse_alpha = .001) ) /
p_t4 /
(get_plot(silvie_MOT01, repel = F, min_freq =3, ellipse_alpha = .001) |
get_plot(silvie_MOT02, repel = F, min_freq =3, ellipse_alpha = .001) |
get_plot(silvie_MOT03, repel = F, min_freq =3, ellipse_alpha = .001)) +
plot_layout(heights = c(.1, 1, .1, 1, .1, 1, .1, 1))
# ggsave("images/networks_cds.png", width = 20, height = 30)Largest communities
# networks spanning the entire time period
fion_FAT <- bigrams_cds_fion %>% filter(Speaker == "FAT") %>% get_network_cds()
fion_MOT <- bigrams_cds_fion %>% filter(Speaker == "MOT") %>% get_network_cds()
silvie_FAT <- bigrams_cds_silvie %>% filter(Speaker == "FAT") %>% get_network_cds()
silvie_MOT <- bigrams_cds_silvie %>% filter(Speaker == "MOT") %>% get_network_cds()
# get largest communities
# Silvie's mother
silvie_MOT_network <- silvie_MOT %>% create_layout(layout = "fr")
largest_clusters <- silvie_MOT_network %>% select(community) %>% table() %>% sort(decreasing = T) %>% head(4) %>% enframe() %>% select(name) %>% as_vector() %>% unname()
largest_clusters_prop <- silvie_MOT_network %>% select(community) %>% table() %>% sort(decreasing = T) %>% prop.table() %>% head(4) %>% round(digits = 4)
# plot four largest communities
silvie_MOT_lc <- ( silvie_MOT_network %>% filter(community %in% largest_clusters[1]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 1 (", largest_clusters_prop[1]*100, "%)")) |
silvie_MOT_network %>% filter(community %in% largest_clusters[2]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 2 (", largest_clusters_prop[2]*100, "%)")) |
silvie_MOT_network %>% filter(community %in% largest_clusters[3]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 3 (", largest_clusters_prop[3]*100, "%)")) |
silvie_MOT_network %>% filter(community %in% largest_clusters[4]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 4 (", largest_clusters_prop[4]*100, "%)")))
# Silvie's father
silvie_FAT_network <- silvie_FAT %>% create_layout(layout = "fr")
largest_clusters <- silvie_FAT_network %>% select(community) %>% table() %>% sort(decreasing = T) %>% head(4) %>% enframe() %>% select(name) %>% as_vector() %>% unname()
largest_clusters_prop <- silvie_FAT_network %>% select(community) %>% table() %>% sort(decreasing = T) %>% prop.table() %>% head(4) %>% round(digits = 4)
# plot four largest communities
silvie_FAT_lc <- ( silvie_FAT_network %>% filter(community %in% largest_clusters[1]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 1 (", largest_clusters_prop[1]*100, "%)")) |
silvie_FAT_network %>% filter(community %in% largest_clusters[2]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 2 (", largest_clusters_prop[2]*100, "%)")) |
silvie_FAT_network %>% filter(community %in% largest_clusters[3]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 3 (", largest_clusters_prop[3]*100, "%)")) |
silvie_FAT_network %>% filter(community %in% largest_clusters[4]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 4 (", largest_clusters_prop[4]*100, "%)")))
# Fion's mother
fion_MOT_network <- fion_MOT %>% create_layout(layout = "fr")
largest_clusters <- fion_MOT_network %>% select(community) %>% table() %>% sort(decreasing = T) %>% head(4) %>% enframe() %>% select(name) %>% as_vector() %>% unname()
largest_clusters_prop <- fion_MOT_network %>% select(community) %>% table() %>% sort(decreasing = T) %>% prop.table() %>% head(4) %>% round(digits = 4)
# plot four largest communities
fion_MOT_lc <- ( fion_MOT_network %>% filter(community %in% largest_clusters[1]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 1 (", largest_clusters_prop[1]*100, "%)")) |
fion_MOT_network %>% filter(community %in% largest_clusters[2]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 2 (", largest_clusters_prop[2]*100, "%)")) |
fion_MOT_network %>% filter(community %in% largest_clusters[3]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 3 (", largest_clusters_prop[3]*100, "%)")) |
fion_MOT_network %>% filter(community %in% largest_clusters[4]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 4 (", largest_clusters_prop[4]*100, "%)")))
# fion's father
fion_FAT_network <- fion_FAT %>% create_layout(layout = "fr")
largest_clusters <- fion_FAT_network %>% select(community) %>% table() %>% sort(decreasing = T) %>% head(4) %>% enframe() %>% select(name) %>% as_vector() %>% unname()
largest_clusters_prop <- fion_FAT_network %>% select(community) %>% table() %>% sort(decreasing = T) %>% prop.table() %>% head(4) %>% round(digits = 4)
# plot four largest communities
fion_FAT_lc <- ( fion_FAT_network %>% filter(community %in% largest_clusters[1]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 1 (", largest_clusters_prop[1]*100, "%)")) |
fion_FAT_network %>% filter(community %in% largest_clusters[2]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 2 (", largest_clusters_prop[2]*100, "%)")) |
fion_FAT_network %>% filter(community %in% largest_clusters[3]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 3 (", largest_clusters_prop[3]*100, "%)")) |
fion_FAT_network %>% filter(community %in% largest_clusters[4]) %>% get_plot(input = "layout") + ggtitle(paste0("Community 4 (", largest_clusters_prop[4]*100, "%)")))
# all in one big plot
title_fion_MOT <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Fion', ' MOT'))", size = 12, parse = TRUE) + theme_void()
title_fion_FAT <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Fion', ' FAT'))", size = 12, parse = TRUE) + theme_void()
title_silvie_MOT <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Silvie', ' MOT'))", size = 12, parse = TRUE) + theme_void()
title_silvie_FAT <- ggplot() + annotate("text", x = 4, y = 25, label = "bold(paste('Silvie', ' FAT'))", size = 12, parse = TRUE) + theme_void() Plots for online viewing
Here we provide plots showing the full networks for each time slice for online viewing.
Child speech
# Fion
bigrams_fion01 %>% get_network() %>% get_plot(repel = FALSE, interactive = TRUE)bigrams_fion02 %>% get_network() %>% get_plot(repel = FALSE,interactive = TRUE)bigrams_fion03 %>% get_network() %>% get_plot(repel = FALSE,interactive = TRUE)bigrams_fion04 %>% get_network() %>% get_plot(repel = FALSE,interactive = TRUE)bigrams_fion05 %>% get_network() %>% get_plot(repel = FALSE,interactive = TRUE)bigrams_fion06 %>% get_network() %>% get_plot(repel = FALSE,interactive = TRUE)bigrams_fion07 %>% get_network() %>% get_plot(repel = FALSE,interactive = TRUE)# Silvie
bigrams_silvie01 %>% get_network() %>% get_plot(repel = FALSE, interactive = TRUE)bigrams_silvie02 %>% get_network() %>% get_plot(repel = FALSE,interactive = TRUE)bigrams_silvie03 %>% get_network() %>% get_plot(repel = FALSE,interactive = TRUE)bigrams_silvie04 %>% get_network() %>% get_plot(repel = FALSE,interactive = TRUE)bigrams_silvie05 %>% get_network() %>% get_plot(repel = FALSE,interactive = TRUE)bigrams_silvie06 %>% get_network() %>% get_plot(repel = FALSE,interactive = TRUE)Child-directed speech
Network size
Network size refers to the number of nodes in the network. We calculate it for each child in each month.
Child speech
network_sizes <- bind_rows(
tibble(
Child = "Fion",
Language = "German",
Size = sapply(1:length(levels(factor(bigrams_fion$Month))), function(i) length(V(bigrams_fion %>% filter(Month == levels(factor(bigrams_fion$Month))[i] & type == "german") %>% select(word1, word2) %>% graph_from_data_frame()))),
Month = levels(factor(bigrams_fion$Month))
),
tibble(
Child = "Fion",
Language = "English",
Size = sapply(1:length(levels(factor(bigrams_fion$Month))), function(i) length(V(bigrams_fion %>% filter(Month == levels(factor(bigrams_fion$Month))[i] & type == "english") %>% select(word1, word2) %>% graph_from_data_frame()))),
Month = levels(factor(bigrams_fion$Month))
),
tibble(
Child = "Fion",
Language = "mixed",
Size = sapply(1:length(levels(factor(bigrams_fion$Month))), function(i) length(V(bigrams_fion %>% filter(Month == levels(factor(bigrams_fion$Month))[i] & type == "mixed") %>% select(word1, word2) %>% graph_from_data_frame()))),
Month = levels(factor(bigrams_fion$Month))
),
tibble(
Child = "Fion",
Language = "all",
Size = sapply(1:length(levels(factor(bigrams_fion$Month))), function(i) length(V(bigrams_fion %>% filter(Month == levels(factor(bigrams_fion$Month))[i]) %>% select(word1, word2) %>% graph_from_data_frame()))),
Month = levels(factor(bigrams_fion$Month))
),
tibble(
Child = "Silvie",
Language = "German",
Size = sapply(1:length(levels(factor(bigrams_silvie$Month))), function(i) length(V(bigrams_silvie %>% filter(Month == levels(factor(bigrams_silvie$Month))[i] & type == "german") %>% select(word1, word2) %>% graph_from_data_frame()))),
Month = levels(factor(bigrams_silvie$Month))
),
tibble(
Child = "Silvie",
Language = "English",
Size = sapply(1:length(levels(factor(bigrams_silvie$Month))), function(i) length(V(bigrams_silvie %>% filter(Month == levels(factor(bigrams_silvie$Month))[i] & type == "english") %>% select(word1, word2) %>% graph_from_data_frame()))),
Month = levels(factor(bigrams_silvie$Month))
),
tibble(
Child = "Silvie",
Language = "mixed",
Size = sapply(1:length(levels(factor(bigrams_silvie$Month))), function(i) length(V(bigrams_silvie %>% filter(Month == levels(factor(bigrams_silvie$Month))[i] & type == "mixed") %>% select(word1, word2) %>% graph_from_data_frame()))),
Month = levels(factor(bigrams_silvie$Month))
),
tibble(
Child = "Silvie",
Language = "all",
Size = sapply(1:length(levels(factor(bigrams_silvie$Month))), function(i) length(V(bigrams_silvie %>% filter(Month == levels(factor(bigrams_silvie$Month))[i]) %>% select(word1, word2) %>% graph_from_data_frame()))),
Month = levels(factor(bigrams_silvie$Month))
)
)
# add columns for letting the "all" line appear differently
network_sizes$alpha <- ifelse(network_sizes$Language=="all", 0.3, 1)
network_sizes$linetype <- ifelse(network_sizes$Language == "all", 2, 1)
network_sizes$linewidth <- ifelse(network_sizes$Language == "all", 2.5, .8)
ggplot(network_sizes, aes(x = Month, y = Size, group = Language, col = Language, linewidth = linewidth, linetype = linetype, alpha = alpha)) + geom_point() + geom_line() + facet_wrap(~Child) + theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
scale_color_manual(values = c("darkgreen", "deepskyblue", "salmon", "tan")) + scale_alpha_identity() + scale_linetype_identity() + scale_linewidth_identity() +
theme(axis.text = element_text(size = 18)) +
theme(axis.title = element_text(size = 18)) +
theme(strip.text = element_text(size = 18)) +
theme(legend.text = element_text(size = 18)) +
theme(legend.title = element_text(size = 18, face = "bold")) +
theme(text = element_text(size = 18))
# network sizes for the sampled data
network_sizes_sampled <- bind_rows(
tibble(
Child = "Fion",
Language = "German",
Period = levels(factor(bigrams_fion$age_range)),
Size = sapply(1:7, function(i) length(V(get(paste0("bigrams_fion0", i)) %>% filter(type == "german") %>% select(word1, word2) %>% graph_from_data_frame())))
),
tibble(
Child = "Fion",
Language = "English",
Period = levels(factor(bigrams_fion$age_range)),
Size = sapply(1:7, function(i) length(V(get(paste0("bigrams_fion0", i)) %>% filter(type == "english") %>% select(word1, word2) %>% graph_from_data_frame())))
),
tibble(
Child = "Fion",
Language = "mixed",
Period = levels(factor(bigrams_fion$age_range)),
Size = sapply(1:7, function(i) length(V(get(paste0("bigrams_fion0", i)) %>% filter(type == "mixed") %>% select(word1, word2) %>% graph_from_data_frame())))
),
tibble(
Child = "Fion",
Language = "all",
Period = levels(factor(bigrams_fion$age_range)),
Size = sapply(1:7, function(i) length(V(get(paste0("bigrams_fion0", i)) %>% select(word1, word2) %>% graph_from_data_frame())))
),
tibble(
Child = "Silvie",
Language = "German",
Period = levels(factor(bigrams_silvie$age_range)),
Size = sapply(1:6, function(i) length(V(get(paste0("bigrams_silvie0", i)) %>% filter(type == "german") %>% select(word1, word2) %>% graph_from_data_frame())))
),
tibble(
Child = "Silvie",
Language = "English",
Period = levels(factor(bigrams_silvie$age_range)),
Size = sapply(1:6, function(i) length(V(get(paste0("bigrams_silvie0", i)) %>% filter(type == "english") %>% select(word1, word2) %>% graph_from_data_frame())))
),
tibble(
Child = "Silvie",
Language = "mixed",
Period = levels(factor(bigrams_silvie$age_range)),
Size = sapply(1:6, function(i) length(V(get(paste0("bigrams_silvie0", i)) %>% filter(type == "mixed") %>% select(word1, word2) %>% graph_from_data_frame())))
),
tibble(
Child = "Silvie",
Language = "all",
Period = levels(factor(bigrams_silvie$age_range)),
Size = sapply(1:6, function(i) length(V(get(paste0("bigrams_silvie0", i)) %>% select(word1, word2) %>% graph_from_data_frame())))
)
)
# add columns for letting the "all" line appear differently
network_sizes_sampled$alpha <- ifelse(network_sizes_sampled$Language=="all", 0.3, 1)
network_sizes_sampled$linetype <- ifelse(network_sizes_sampled$Language == "all", 2, 1)
network_sizes_sampled$linewidth <- ifelse(network_sizes_sampled$Language == "all", 2.5, .8)
ggplot(network_sizes_sampled, aes(x = Period, y = Size, group = Language, col = Language, linewidth = linewidth, linetype = linetype, alpha = alpha)) + geom_point() + geom_line() + facet_wrap(~Child, scales = "free_x") + theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
scale_color_manual(values = c("darkgreen", "deepskyblue", "salmon", "tan")) + scale_alpha_identity() + scale_linetype_identity() + scale_linewidth_identity() +
theme(axis.text = element_text(size = 18)) +
theme(axis.title = element_text(size = 18)) +
theme(strip.text = element_text(size = 18)) +
theme(legend.text = element_text(size = 18)) +
theme(legend.title = element_text(size = 18, face = "bold")) +
theme(text = element_text(size = 18))
# ggsave("images/network_size.png", width = 12, height = 6) Child-directed speech
network_sizes_cds <- bind_rows(
tibble(
Speaker = "FAT",
Child = "Silvie",
Language = "English",
Period = levels(factor(bigrams_cds_silvie$age_range)),
`Number of nodes` = sapply(1:3, function(i) length(V(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="FAT" & lang == "en") %>% get_network_cds()))),
`Number of edges` = sapply(1:3, function(i) length(E(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="FAT" & lang == "en") %>% get_network_cds()))),
`Average weight` = paste0(sapply(1:3, function(i) mean(E(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="FAT" & lang == "en") %>% get_network_cds())$weight)) %>% round(digits = 3), " (", sapply(1:3, function(i) sd(E(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="FAT" & lang == "en") %>% get_network_cds())$weight)) %>% round(digits = 3), ")")
),
tibble(
Speaker = "FAT",
Child = "Silvie",
Language = "German",
Period = levels(factor(bigrams_cds_silvie$age_range)),
`Number of nodes` = sapply(1:3, function(i) length(V(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="FAT" & lang == "de") %>% get_network_cds()))),
`Number of edges` = sapply(1:3, function(i) length(E(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="FAT" & lang == "de") %>% get_network_cds()))),
`Average weight` = paste0(sapply(1:3, function(i) mean(E(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="FAT" & lang == "de") %>% get_network_cds())$weight)) %>% round(digits = 3), " (", sapply(1:3, function(i) sd(E(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="FAT" & lang == "de") %>% get_network_cds())$weight)) %>% round(digits = 3), ")")
),
tibble(
Speaker = "MOT",
Child = "Silvie",
Language = "English",
Period = levels(factor(bigrams_cds_silvie$age_range)),
`Number of nodes` = sapply(1:3, function(i) length(V(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="MOT" & lang == "en") %>% get_network_cds()))),
`Number of edges` = sapply(1:3, function(i) length(E(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="MOT" & lang == "en") %>% get_network_cds()))),
`Average weight` = paste0(sapply(1:3, function(i) mean(E(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="MOT" & lang == "en") %>% get_network_cds())$weight)) %>% round(digits = 3), " (", sapply(1:3, function(i) sd(E(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="MOT" & lang == "en") %>% get_network_cds())$weight)) %>% round(digits = 3), ")")
),
tibble(
Speaker = "MOT",
Child = "Silvie",
Language = "German",
Period = levels(factor(bigrams_cds_silvie$age_range)),
`Number of nodes` = sapply(1:3, function(i) length(V(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="MOT" & lang == "de") %>% get_network_cds()))),
`Number of edges` = sapply(1:3, function(i) length(E(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="MOT" & lang == "de") %>% get_network_cds()))),
`Average weight` = paste0(sapply(1:3, function(i) mean(E(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="MOT" & lang == "de") %>% get_network_cds())$weight)) %>% round(digits = 3), " (", sapply(1:3, function(i) sd(E(get(paste0("bigrams_cds_silvie0", i)) %>% filter(Speaker=="MOT" & lang == "de") %>% get_network_cds())$weight)) %>% round(digits = 3), ")")
),
tibble(
Speaker = "FAT",
Child = "Fion",
Language = "English",
Period = levels(factor(bigrams_cds_fion$age_range)),
`Number of nodes` = sapply(1:3, function(i) length(V(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="FAT" & lang == "en") %>% get_network_cds()))),
`Number of edges` = sapply(1:3, function(i) length(E(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="FAT" & lang == "en") %>% get_network_cds()))),
`Average weight` = paste0(sapply(1:3, function(i) mean(E(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="FAT" & lang == "en") %>% get_network_cds())$weight)) %>% round(digits = 3), " (", sapply(1:3, function(i) sd(E(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="FAT" & lang == "en") %>% get_network_cds())$weight)) %>% round(digits = 3), ")")
),
tibble(
Speaker = "FAT",
Child = "Fion",
Language = "German",
Period = levels(factor(bigrams_cds_fion$age_range)),
`Number of nodes` = sapply(1:3, function(i) length(V(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="FAT" & lang == "de") %>% get_network_cds()))),
`Number of edges` = sapply(1:3, function(i) length(E(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="FAT" & lang == "de") %>% get_network_cds()))),
`Average weight` = paste0(sapply(1:3, function(i) mean(E(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="FAT" & lang == "de") %>% get_network_cds())$weight)) %>% round(digits = 3), " (", sapply(1:3, function(i) sd(E(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="FAT" & lang == "de") %>% get_network_cds())$weight)) %>% round(digits = 3), ")")
),
tibble(
Speaker = "MOT",
Child = "Fion",
Language = "English",
Period = levels(factor(bigrams_cds_fion$age_range)),
`Number of nodes` = sapply(1:3, function(i) length(V(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="MOT" & lang == "en") %>% get_network_cds()))),
`Number of edges` = sapply(1:3, function(i) length(E(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="MOT" & lang == "en") %>% get_network_cds()))),
`Average weight` = paste0(sapply(1:3, function(i) mean(E(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="MOT" & lang == "en") %>% get_network_cds())$weight)) %>% round(digits = 3), " (", sapply(1:3, function(i) sd(E(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="MOT" & lang == "en") %>% get_network_cds())$weight)) %>% round(digits = 3), ")")
),
tibble(
Speaker = "MOT",
Child = "Fion",
Language = "German",
Period = levels(factor(bigrams_cds_fion$age_range)),
`Number of nodes` = sapply(1:3, function(i) length(V(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="MOT" & lang == "de") %>% get_network_cds()))),
`Number of edges` = sapply(1:3, function(i) length(E(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="MOT" & lang == "de") %>% get_network_cds()))),
`Average weight` = paste0(sapply(1:3, function(i) mean(E(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="MOT" & lang == "de") %>% get_network_cds())$weight)) %>% round(digits = 3), " (", sapply(1:3, function(i) sd(E(get(paste0("bigrams_cds_fion0", i)) %>% filter(Speaker=="MOT" & lang == "de") %>% get_network_cds())$weight)) %>% round(digits = 3), ")")
)
)
# pivot to wider table
network_sizes_cds %>% pivot_wider(names_from = "Period", values_from = c("Number of nodes", "Number of edges", "Average weight")) #%>% writexl::write_xlsx("data/cds_network_sizes.xlsx")# A tibble: 8 × 21
Speaker Child Language `Number of nodes_02;04-02;10` Number of nodes_02;11-…¹
<chr> <chr> <chr> <int> <int>
1 FAT Silvie English 359 350
2 FAT Silvie German 1668 1548
3 MOT Silvie English 2726 2548
4 MOT Silvie German 636 633
5 FAT Fion English NA NA
6 FAT Fion German NA NA
7 MOT Fion English NA NA
8 MOT Fion German NA NA
# ℹ abbreviated name: ¹`Number of nodes_02;11-03;04`
# ℹ 16 more variables: `Number of nodes_03;05-03;09` <int>,
# `Number of nodes_02;03-02;10` <int>, `Number of nodes_02;11-03;05` <int>,
# `Number of nodes_03;06-03;11` <int>, `Number of edges_02;04-02;10` <int>,
# `Number of edges_02;11-03;04` <int>, `Number of edges_03;05-03;09` <int>,
# `Number of edges_02;03-02;10` <int>, `Number of edges_02;11-03;05` <int>,
# `Number of edges_03;06-03;11` <int>, `Average weight_02;04-02;10` <chr>, …Average degree of networks
We are interested in the average degree of the networks, i.e. the total number of links divided by the total number of nodes. The average degree of a network can be seen as an index of the “productivity” of the words in use (here to be understood as the extent to which they combine with other words and appear in different contexts) and of the complexity of the utterances in question.
Here it is particularly interesting to take a closer look at the average degree of the German and English utterances in isolation, as well as the entire network (which also contains code-mixed utterances).
fion_network_DE <- bigrams_fion %>% filter(type == "german") %>% get_network()
fion_network_EN <- bigrams_fion %>% filter(type == "english") %>% get_network()
# get a dataframe with degree scores for each month
# for each child and alllanguage types (English, German, mixed, all)
degree_tbl <- bind_rows(
tibble(Month = levels(factor(bigrams_fion$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_fion$Month))), function(i) mean(degree(bigrams_fion %>% filter(type == "german" & Month == levels(factor(bigrams_fion$Month))[i]) %>% get_network()))),
Child = "Fion",
Language = "German"),
tibble(Month = levels(factor(bigrams_fion$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_fion$Month))), function(i) mean(degree(bigrams_fion %>% filter(type == "english" & Month == levels(factor(bigrams_fion$Month))[i]) %>% get_network()))),
Child = "Fion",
Language = "English"),
tibble(Month = levels(factor(bigrams_silvie$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_silvie$Month))), function(i) mean(degree(bigrams_silvie %>% filter(type == "german" & Month == levels(factor(bigrams_silvie$Month))[i]) %>% get_network()))),
Child = "Silvie",
Language = "German"),
tibble(Month = levels(factor(bigrams_silvie$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_silvie$Month))), function(i) mean(degree(bigrams_silvie %>% filter(type == "english" & Month == levels(factor(bigrams_silvie$Month))[i]) %>% get_network()))),
Child = "Silvie",
Language = "English"),
tibble(Month = levels(factor(bigrams_fion$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_fion$Month))), function(i) mean(degree(bigrams_fion %>% filter(type == "mixed" & Month == levels(factor(bigrams_fion$Month))[i]) %>% get_network()))),
Child = "Fion",
Language = "mixed"),
tibble(Month = levels(factor(bigrams_silvie$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_silvie$Month))), function(i) mean(degree(bigrams_silvie %>% filter(type == "mixed" & Month == levels(factor(bigrams_silvie$Month))[i]) %>% get_network()))),
Child = "Silvie",
Language = "mixed"),
tibble(Month = levels(factor(bigrams_fion$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_fion$Month))), function(i) mean(degree(bigrams_fion %>% filter(Month == levels(factor(bigrams_fion$Month))[i]) %>% get_network()))),
Child = "Fion",
Language = "all"),
tibble(Month = levels(factor(bigrams_silvie$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_silvie$Month))), function(i) mean(degree(bigrams_silvie %>% filter(Month == levels(factor(bigrams_silvie$Month))[i]) %>% get_network()))),
Child = "Silvie",
Language = "all")
)
# add columns for letting the "all" line appear differently
degree_tbl$alpha <- ifelse(degree_tbl$Language=="all", 0.3, 1)
degree_tbl$linetype <- ifelse(degree_tbl$Language == "all", 2, 1)
degree_tbl$linewidth <- ifelse(degree_tbl$Language == "all", 2.5, .8)
degree_tbl %>% ggplot(aes(x = Month, y = `Average degree`, group = Language, col = Language, alpha = alpha, linetype = linetype, linewidth = linewidth)) +
geom_point() + geom_line() + facet_wrap(~Child) +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
scale_color_manual(values = c("darkgreen", "deepskyblue", "salmon", "tan")) + scale_alpha_identity() + scale_linetype_identity() + scale_linewidth_identity() +
theme(axis.text = element_text(size = 18)) +
theme(axis.title = element_text(size = 18)) +
theme(strip.text = element_text(size = 18)) +
theme(legend.text = element_text(size = 18)) +
theme(legend.title = element_text(size = 18, face = "bold")) +
theme(text = element_text(size = 18))
# ggsave("images/average_degree_fion_silvie.png", width = 12, height = 6)
# Child-directed speech
degree_tbl_cds <- bind_rows(
tibble(
Month = levels(factor(bigrams_cds_fion$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_cds_fion$Month))), function(i) mean(degree(bigrams_cds_fion %>% filter(Speaker == "FAT" & lang == "en" & Month == levels(factor(bigrams_cds_fion$Month))[i]) %>% get_network_cds()))),
Speaker = "Fion's father",
Language = "English"
),
tibble(
Month = levels(factor(bigrams_cds_fion$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_cds_fion$Month))), function(i) mean(degree(bigrams_cds_fion %>% filter(Speaker == "FAT" & lang == "de" & Month == levels(factor(bigrams_cds_fion$Month))[i]) %>% get_network_cds()))),
Speaker = "Fion's father",
Language = "German"
),
tibble(
Month = levels(factor(bigrams_cds_fion$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_cds_fion$Month))), function(i) mean(degree(bigrams_cds_fion %>% filter(Speaker == "FAT" & Month == levels(factor(bigrams_cds_fion$Month))[i]) %>% get_network_cds()))),
Speaker = "Fion's father",
Language = "all"
),
tibble(
Month = levels(factor(bigrams_cds_silvie$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_cds_silvie$Month))), function(i) mean(degree(bigrams_cds_silvie %>% filter(Speaker == "FAT" & lang == "en" & Month == levels(factor(bigrams_cds_silvie$Month))[i]) %>% get_network_cds()))),
Speaker = "Silvie's father",
Language = "English"
),
tibble(
Month = levels(factor(bigrams_cds_silvie$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_cds_silvie$Month))), function(i) mean(degree(bigrams_cds_silvie %>% filter(Speaker == "FAT" & lang == "de" & Month == levels(factor(bigrams_cds_silvie$Month))[i]) %>% get_network_cds()))),
Speaker = "Silvie's father",
Language = "German"
),
tibble(
Month = levels(factor(bigrams_cds_silvie$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_cds_silvie$Month))), function(i) mean(degree(bigrams_cds_silvie %>% filter(Speaker == "FAT" & Month == levels(factor(bigrams_cds_silvie$Month))[i]) %>% get_network_cds()))),
Speaker = "Silvie's father",
Language = "all"
),
tibble(
Month = levels(factor(bigrams_cds_fion$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_cds_fion$Month))), function(i) mean(degree(bigrams_cds_fion %>% filter(Speaker == "MOT" & lang == "en" & Month == levels(factor(bigrams_cds_fion$Month))[i]) %>% get_network_cds()))),
Speaker = "Fion's mother",
Language = "English"
),
tibble(
Month = levels(factor(bigrams_cds_fion$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_cds_fion$Month))), function(i) mean(degree(bigrams_cds_fion %>% filter(Speaker == "MOT" & lang == "de" & Month == levels(factor(bigrams_cds_fion$Month))[i]) %>% get_network_cds()))),
Speaker = "Fion's mother",
Language = "German"
),
tibble(
Month = levels(factor(bigrams_cds_fion$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_cds_fion$Month))), function(i) mean(degree(bigrams_cds_fion %>% filter(Speaker == "MOT" & Month == levels(factor(bigrams_cds_fion$Month))[i]) %>% get_network_cds()))),
Speaker = "Fion's mother",
Language = "all"
),
tibble(
Month = levels(factor(bigrams_cds_silvie$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_cds_silvie$Month))), function(i) mean(degree(bigrams_cds_silvie %>% filter(Speaker == "MOT" & lang == "en" & Month == levels(factor(bigrams_cds_silvie$Month))[i]) %>% get_network_cds()))),
Speaker = "Silvie's mother",
Language = "English"
),
tibble(
Month = levels(factor(bigrams_cds_silvie$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_cds_silvie$Month))), function(i) mean(degree(bigrams_cds_silvie %>% filter(Speaker == "MOT" & lang == "de" & Month == levels(factor(bigrams_cds_silvie$Month))[i]) %>% get_network_cds()))),
Speaker = "Silvie's mother",
Language = "German"
),
tibble(
Month = levels(factor(bigrams_cds_silvie$Month)),
`Average degree` = sapply(1:length(levels(factor(bigrams_cds_silvie$Month))), function(i) mean(degree(bigrams_cds_silvie %>% filter(Speaker == "MOT" & Month == levels(factor(bigrams_cds_silvie$Month))[i]) %>% get_network_cds()))),
Speaker = "Silvie's mother",
Language = "all"
)
)
# add columns for letting the "all" line appear differently
degree_tbl_cds$alpha <- ifelse(degree_tbl_cds$Language=="all", 0.3, 1)
degree_tbl_cds$linetype <- ifelse(degree_tbl_cds$Language == "all", 2, 1)
degree_tbl_cds$linewidth <- ifelse(degree_tbl_cds$Language == "all", 2.5, .8)
degree_tbl_cds %>% ggplot(aes(x = Month, y = `Average degree`, group = Language, col = Language, alpha = alpha, linetype = linetype, linewidth = linewidth)) + geom_point() + geom_line() + facet_wrap(~Speaker) +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
scale_color_manual(values = c("darkgreen", "deepskyblue", "salmon")) + scale_alpha_identity() + scale_linetype_identity() + scale_linewidth_identity() +
theme(axis.text = element_text(size = 18)) +
theme(axis.title = element_text(size = 18)) +
theme(strip.text = element_text(size = 18)) +
theme(legend.text = element_text(size = 18)) +
theme(legend.title = element_text(size = 18, face = "bold")) +
theme(text = element_text(size = 18))
# ggsave("images/degree_cds.png", width = 10, height = 10)Distribution of languages in CDS
To better understand the outcomes of the DNM analysis, we check the number of utterances and the distribution of different language tags in the CDS data.
# get number of utterances and average number of words per utterance for each speaker and month
se <- function(x) sqrt(var(x) / length(x))
fion_parents_mlu <- fion_cds %>% filter(Speaker %in% c("MOT", "FAT")) %>% group_by(Speaker, Month, lang) %>% summarise(
mean = mean(Wordcount),
se = se(Wordcount),
se_min = mean-se,
se_max = mean+se
) %>% ggplot(aes(x = Month, y = mean, group = lang, col = lang)) + geom_point() + geom_line() + geom_errorbar(aes(ymin = se_min, ymax = se_max)) + facet_wrap(~Speaker) + ylab("Mean length of utterances") +
scale_color_manual(values = c("salmon", "deepskyblue")) +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
theme(axis.text = element_text(size = 18)) +
theme(axis.title = element_text(size = 18)) +
theme(strip.text = element_text(size = 18)) +
theme(legend.text = element_text(size = 18)) +
theme(legend.title = element_text(size = 18, face = "bold")) +
theme(text = element_text(size = 18)) + ggtitle("Fion's parents") + theme(plot.title = element_text(face = "bold", hjust = 0.5))
silvie_parents_mlu <- silvie_cds %>% filter(Speaker %in% c("MOT", "FAT")) %>% group_by(Speaker, Month, lang) %>% summarise(
mean = mean(Wordcount),
se = se(Wordcount),
se_min = mean-se,
se_max = mean+se
) %>% ggplot(aes(x = Month, y = mean, group = lang, col = lang)) + geom_point() + geom_line() + geom_errorbar(aes(ymin = se_min, ymax = se_max)) + facet_wrap(~Speaker) + ylab("Mean length of utterances") +
scale_color_manual(values = c("salmon", "deepskyblue")) +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
theme(axis.text = element_text(size = 18)) +
theme(axis.title = element_text(size = 18)) +
theme(strip.text = element_text(size = 18)) +
theme(legend.text = element_text(size = 18)) +
theme(legend.title = element_text(size = 18, face = "bold")) +
theme(text = element_text(size = 18)) + ggtitle("Silvie's parents") + theme(plot.title = element_text(face = "bold", hjust = 0.5))
fion_parents_mlu /
silvie_parents_mlu + plot_layout(axis_titles = "collect", guides = "collect")
fion_parents_n <- fion_cds %>% filter(Speaker %in% c("MOT", "FAT")) %>% group_by(Speaker, Month, lang) %>% filter(Speaker %in% c("MOT", "FAT")) %>% group_by(Speaker, Month, lang) %>% summarise(
n = n(),
) %>% mutate(Language = case_when(lang=="de" ~ "German", lang=="en" ~ "English")) %>% ggplot(aes(x = Month, y = n, group = Language, col = Language)) + geom_point() + geom_line() + facet_wrap(~Speaker) + ylab("Number of utterances") +
scale_color_manual(values = c("salmon", "deepskyblue")) +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
theme(axis.text = element_text(size = 18)) +
theme(axis.title = element_text(size = 18)) +
theme(strip.text = element_text(size = 18)) +
theme(legend.text = element_text(size = 18)) +
theme(legend.title = element_text(size = 18, face = "bold")) +
theme(text = element_text(size = 18)) + ggtitle("Fion's parents") + theme(plot.title = element_text(face = "bold", hjust = 0.5))
silvie_parents_n <-silvie_cds %>% filter(Speaker %in% c("MOT", "FAT")) %>% group_by(Speaker, Month, lang) %>% summarise(
n = n(),
) %>% mutate(Language = case_when(lang=="de" ~ "German", lang=="en" ~ "English")) %>% ggplot(aes(x = Month, y = n, group = Language, col = Language)) + geom_point() + geom_line() + facet_wrap(~Speaker) + ylab("Number of utterances") +
scale_color_manual(values = c("salmon", "deepskyblue")) +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
theme(axis.text = element_text(size = 18)) +
theme(axis.title = element_text(size = 18)) +
theme(strip.text = element_text(size = 18)) +
theme(legend.text = element_text(size = 18)) +
theme(legend.title = element_text(size = 18, face = "bold")) +
theme(text = element_text(size = 18)) + ggtitle("Silvie's parents") + theme(plot.title = element_text(face = "bold", hjust = 0.5))
fion_parents_n /
silvie_parents_n + plot_layout(axis_titles = "collect", guides = "collect")
# ggsave("images/cds_number_of_utterances.png", width = 11, height = 9)Modularities
Modularity measures the density of links inside communities compared to links between different communities and is defined as a value between -1 and 1.
Child-directed speech
cds_modularities <- bind_rows(
tibble(
Speaker = "MOT",
Language = "English",
Modularity = sapply(1:length(levels(factor(bigrams_cds_fion$Month))), function(i) bigrams_cds_fion %>% filter(Month == levels(factor(bigrams_cds_fion$Month))[i] & Speaker == "MOT" & lang == "en") %>% get_network_cds() %>% cluster_louvain(resolution = .3) %>% modularity() ),
Month = levels(factor(bigrams_cds_fion$Month)),
Child = "Fion"
),
tibble(
Speaker = "MOT",
Language = "German",
Modularity = sapply(1:length(levels(factor(bigrams_cds_fion$Month))), function(i) bigrams_cds_fion %>% filter(Month == levels(factor(bigrams_cds_fion$Month))[i] & Speaker == "MOT" & lang == "de") %>% get_network_cds() %>% cluster_louvain(resolution = .3) %>% modularity() ),
Month = levels(factor(bigrams_cds_fion$Month)),
Child = "Fion"
),
tibble(
Speaker = "FAT",
Language = "English",
Modularity = sapply(1:length(levels(factor(bigrams_cds_fion$Month))), function(i) bigrams_cds_fion %>% filter(Month == levels(factor(bigrams_cds_fion$Month))[i] & Speaker == "FAT" & lang=="en") %>% get_network_cds() %>% cluster_louvain(resolution = .3) %>% modularity() ),
Month = levels(factor(bigrams_cds_fion$Month)),
Child = "Fion"
),
tibble(
Speaker = "FAT",
Language = "German",
Modularity = sapply(1:length(levels(factor(bigrams_cds_fion$Month))), function(i) bigrams_cds_fion %>% filter(Month == levels(factor(bigrams_cds_fion$Month))[i] & Speaker == "FAT" & lang=="de") %>% get_network_cds() %>% cluster_louvain(resolution = .3) %>% modularity() ),
Month = levels(factor(bigrams_cds_fion$Month)),
Child = "Fion"
),
tibble(
Speaker = "MOT",
Language = "English",
Modularity = sapply(1:length(levels(factor(bigrams_cds_silvie$Month))), function(i) bigrams_cds_silvie %>% filter(Month == levels(factor(bigrams_cds_silvie$Month))[i] & Speaker == "MOT" & lang=="en") %>% get_network_cds() %>% cluster_louvain(resolution = .3) %>% modularity() ),
Month = levels(factor(bigrams_cds_silvie$Month)),
Child = "Silvie"
),
tibble(
Speaker = "MOT",
Language = "German",
Modularity = sapply(1:length(levels(factor(bigrams_cds_silvie$Month))), function(i) bigrams_cds_silvie %>% filter(Month == levels(factor(bigrams_cds_silvie$Month))[i] & Speaker == "MOT" & lang=="de") %>% get_network_cds() %>% cluster_louvain(resolution = .3) %>% modularity() ),
Month = levels(factor(bigrams_cds_silvie$Month)),
Child = "Silvie"
),
tibble(
Speaker = "FAT",
Language = "English",
Modularity = sapply(1:length(levels(factor(bigrams_cds_silvie$Month))), function(i) bigrams_cds_silvie %>% filter(Month == levels(factor(bigrams_cds_silvie$Month))[i] & Speaker == "FAT" & lang == "en") %>% get_network_cds() %>% cluster_louvain(resolution = .3) %>% modularity() ),
Month = levels(factor(bigrams_cds_silvie$Month)),
Child = "Silvie"
),
tibble(
Speaker = "FAT",
Language = "German",
Modularity = sapply(1:length(levels(factor(bigrams_cds_silvie$Month))), function(i) bigrams_cds_silvie %>% filter(Month == levels(factor(bigrams_cds_silvie$Month))[i] & Speaker == "FAT" & lang == "de") %>% get_network_cds() %>% cluster_louvain(resolution = .3) %>% modularity() ),
Month = levels(factor(bigrams_cds_silvie$Month)),
Child = "Silvie"
)
)
cds_modularities %>% ggplot(aes(x = Month, y = Modularity, group = Language, col = Language)) + geom_point() + geom_line() + facet_grid(Child~Speaker, scales = "free_x") +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
theme(axis.text = element_text(size = 18)) +
theme(axis.title = element_text(size = 18)) +
theme(strip.text = element_text(size = 18)) +
theme(legend.text = element_text(size = 18)) +
theme(legend.title = element_text(size = 18, face = "bold")) +
theme(text = element_text(size = 18))
# ggsave("images/cds_modularities.png", width = 12, height = 10)
# does the outlier in the mother's data have to do with data sparsity? - apparently not!
# fion_cds %>% group_by(Speaker, Month) %>% summarise(
# n=n()
# ) %>% filter(Speaker %in% c("MOT", "FAT")) %>%
# ggplot(aes(x = Month, y = n, group = Speaker, col = Speaker)) +
# geom_point() + geom_line()
#
# bigrams_cds_fion %>% filter(Month == "03;08") %>% get_network_cds() %>% get_plot(interactive = T)
# bigrams_cds_fion %>% filter(Month == "03;09") %>% get_network_cds() %>% get_plot()
# bigrams_cds_fion %>% filter(Month == "03;10") %>% get_network_cds() %>% get_plot(interactive = T)Child speech
# modularities over time - original data
bind_rows(
tibble(
modularity = sapply(1:length(levels(factor(bigrams_fion$Month))),
function(i) filter(bigrams_fion, Month == levels(factor(bigrams_fion$Month))[i]) %>% get_network() %>% cluster_louvain(resolution = .3) %>% modularity()),
age = levels(factor(bigrams_fion$Month)),
Child = "Fion"
),
tibble(
modularity = sapply(1:length(levels(factor(bigrams_silvie$Month))),
function(i) filter(bigrams_silvie, Month == levels(factor(bigrams_silvie$Month))[i]) %>% get_network() %>% cluster_louvain(resolution = .3) %>% modularity()),
age = levels(factor(bigrams_silvie$Month)),
Child = "Silvie"
)
) %>% ggplot(aes(x = age, y = modularity, group = Child, col = Child, pch = Child)) + geom_point() + geom_line() +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
geom_smooth(linewidth = 0) + ylab("Modularity") + xlab("Age")
# Modularities over time - sampled data
modularities_sampled <- bind_rows(
tibble(
Period = levels(factor(bigrams_fion$age_range)),
Modularity = sapply(1:7, function(i) filter(get(paste0("bigrams_fion0", i)), type == "german") %>% get_network() %>% cluster_louvain(resolution = .3) %>% modularity()),
Child = "Fion",
Language = "German"
),
tibble(
Period = levels(factor(bigrams_fion$age_range)),
Modularity = sapply(1:7, function(i) filter(get(paste0("bigrams_fion0", i)), type == "english") %>% get_network() %>% cluster_louvain(resolution = .3) %>% modularity()),
Child = "Fion",
Language = "English"
),
tibble(
Period = levels(factor(bigrams_fion$age_range)),
Modularity = sapply(1:7, function(i) filter(get(paste0("bigrams_fion0", i)), type == "mixed") %>% get_network() %>% cluster_louvain(resolution = .3) %>% modularity()),
Child = "Fion",
Language = "mixed"
),
tibble(
Period = levels(factor(bigrams_fion$age_range)),
Modularity = sapply(1:7, function(i) filter(get(paste0("bigrams_fion0", i))) %>% get_network() %>% cluster_louvain(resolution = .3) %>% modularity()),
Child = "Fion",
Language = "all"
),
tibble(
Period = levels(factor(bigrams_silvie$age_range)),
Modularity = sapply(1:6, function(i) filter(get(paste0("bigrams_silvie0", i)), type == "german") %>% get_network() %>% cluster_louvain(resolution = .3) %>% modularity()),
Child = "Silvie",
Language = "German"
),
tibble(
Period = levels(factor(bigrams_silvie$age_range)),
Modularity = sapply(1:6, function(i) filter(get(paste0("bigrams_silvie0", i)), type == "english") %>% get_network() %>% cluster_louvain(resolution = .3) %>% modularity()),
Child = "Silvie",
Language = "English"
),
tibble(
Period = levels(factor(bigrams_silvie$age_range)),
Modularity = sapply(1:6, function(i) filter(get(paste0("bigrams_silvie0", i)), type == "mixed") %>% get_network() %>% cluster_louvain(resolution = .3) %>% modularity()),
Child = "Silvie",
Language = "mixed"
),
tibble(
Period = levels(factor(bigrams_silvie$age_range)),
Modularity = sapply(1:6, function(i) filter(get(paste0("bigrams_silvie0", i))) %>% get_network() %>% cluster_louvain(resolution = .3) %>% modularity()),
Child = "Silvie",
Language = "all"
)
)
# add columns for letting the "all" line appear differently
modularities_sampled$alpha <- ifelse(modularities_sampled$Language=="all", 0.3, 1)
modularities_sampled$linetype <- ifelse(modularities_sampled$Language == "all", 2, 1)
modularities_sampled$linewidth <- ifelse(modularities_sampled$Language == "all", 2.5, .8)
modularities_sampled %>%
ggplot(aes(x = Period, y = Modularity, group = Language, color = Language, linewidth = linewidth, linetype = linetype, alpha = alpha)) + geom_point() + geom_line() + facet_wrap(~Child, scales = "free_x") +
scale_linetype_identity() +
scale_alpha_identity() +
scale_linewidth_identity() +
scale_color_manual(values = c("darkgreen", "deepskyblue", "salmon", "tan")) +
theme(axis.text.x = element_text(angle=45, hjust=.9, size=12)) +
theme(axis.text = element_text(size = 18)) +
theme(axis.title = element_text(size = 18)) +
theme(strip.text = element_text(size = 18)) +
theme(legend.text = element_text(size = 18)) +
theme(legend.title = element_text(size = 18, face = "bold")) +
theme(text = element_text(size = 18)) +
guides(alpha = "none", linewidth = "none")
# ggsave("images/modularities450.png", width = 12, height = 6)
# What happens with Fion's English data in the 3rd period?
fionp3 <- bigrams_fion %>% filter(age_range == levels(factor(bigrams_fion$age_range))[3] & type == "english") %>% get_network() %>% get_plot() + ggtitle(paste0("Fion's English data, ", levels(factor(bigrams_fion$age_range))[3])) +
theme(plot.title = element_text(face = "bold", hjust = 0.5))
# compare this to the fourth period:
fionp4 <- bigrams_fion %>% filter(age_range == levels(factor(bigrams_fion$age_range))[4] & type == "english") %>% get_network() %>% get_plot() + ggtitle(paste0("Fion's English data, ", levels(factor(bigrams_fion$age_range))[4])) +
theme(plot.title = element_text(face = "bold", hjust = 0.5))
(fionp3 | fionp4)
# ggsave("images/fion_english_data_3and4_450.png", width = 12, height = 5)Number of communities
How many communities are identified?
# communities divided by network size:
sapply(1:7, function(i) get(paste0("network_fion0", i)) %>% cluster_louvain(resolution = .3) %>% communities() %>% length()) / sapply(1:7, function(i) length(V(get(paste0("network_fion0", i)))))[1] 0.17374517 0.13513514 0.10859729 0.13223140 0.11063830 0.08032129 0.09405941sapply(1:6, function(i) get(paste0("network_silvie0", i)) %>% cluster_louvain(resolution = .3) %>% communities() %>% length()) / sapply(1:6, function(i) length(V(get(paste0("network_silvie0", i)))))[1] 0.1254480 0.1553398 0.1344262 0.1147541 0.1025641 0.1317568# relatively stableData overview
bind_rows(
d_fion %>% group_by(age_range) %>% summarise(
`Number of utterances` = n(),
`Number of words` = sum(Wordcount),
Child = "Fion"
),
d_silvie %>% group_by(age_range) %>% summarise(
`Number of utterances` = n(),
`Number of words` = sum(Wordcount),
Child = "Silvie"
)
) # %>% writexl::write_xlsx("data_overview.xlsx")# A tibble: 13 × 4
age_range `Number of utterances` `Number of words` Child
<chr> <int> <dbl> <chr>
1 02;03-02;06 11573 23406 Fion
2 02;07-02;09 11733 30854 Fion
3 02;10-03;00 5572 16548 Fion
4 03;01-03;03 8454 27381 Fion
5 03;04-03;06 4335 14541 Fion
6 03;07-03;09 5075 16321 Fion
7 03;10-03;11 1070 3322 Fion
8 02;04-02;07 10990 31374 Silvie
9 02;08-02;10 7879 25511 Silvie
10 02;11-03;01 7326 25008 Silvie
11 03;02-03;04 6456 23737 Silvie
12 03;05-03;07 2750 10585 Silvie
13 03;08-03;09 2594 10119 Silviebind_rows(
fion_cds %>% group_by(age_range) %>% summarise(
`Number of utterances` = n(),
`Number of words` = sum(Wordcount),
Child = "Fion"
),
silvie_cds %>% group_by(age_range) %>% summarise(
`Number of utterances` = n(),
`Number of words` = sum(Wordcount),
Child = "Silvie"
)
) #%>% writexl::write_xlsx("data_overview_cds.xlsx")# A tibble: 6 × 4
age_range `Number of utterances` `Number of words` Child
<chr> <int> <dbl> <chr>
1 02;03-02;10 48566 173044 Fion
2 02;11-03;05 40048 153388 Fion
3 03;06-03;11 17017 63726 Fion
4 02;04-02;10 49317 268818 Silvie
5 02;11-03;04 31504 167093 Silvie
6 03;05-03;09 13833 72778 Silvie# only parents:
bind_rows(
fion_cds %>% filter(Speaker %in% c("MOT", "FAT")) %>%
group_by(age_range) %>% summarise(
`Number of utterances` = n(),
`Number of words` = sum(Wordcount),
Child = "Fion"
),
silvie_cds %>% filter(Speaker %in% c("MOT", "FAT")) %>%
group_by(age_range) %>% summarise(
`Number of utterances` = n(),
`Number of words` = sum(Wordcount),
Child = "Silvie"
)
)# A tibble: 6 × 4
age_range `Number of utterances` `Number of words` Child
<chr> <int> <dbl> <chr>
1 02;03-02;10 39053 141929 Fion
2 02;11-03;05 32538 129368 Fion
3 03;06-03;11 14684 56672 Fion
4 02;04-02;10 44882 246580 Silvie
5 02;11-03;04 29348 156610 Silvie
6 03;05-03;09 11506 61605 SilvieReferences
Ibbotson, Paul, Vsevolod Salnikov & Richard Walker. 2019. A dynamic network analysis of emergent grammar. First Language 39(6). 652–680. https://doi.org/10.1177/0142723719869562.
Koch, Nikolas, Antje Endesfelder Quick & Stefan Hartmann. 2025. Recycling constructional patterns: The role of chunks in early bilingual acquisition. International Journal of Bilingualism 13670069251346103. https://doi.org/10.1177/13670069251346103.
Quick, Antje Endesfelder, Elena Lieven, Ad Backus & Michael Tomasello. 2018. Constructively combining languages: The use of code-mixing in German-English bilingual child language acquisition. Linguistic Approaches to Bilingualism 8(3). 393–409. https://doi.org/10.1075/lab.17008.qui.